Welcome!

Narrators

TL;DR; The Choose a voice dropdown in the voice dialog lists the narrators available for the voice type you selected. Pick one, hit Confirm, and let the first test video tell you if it's right. Swap if not.

What a narrator is in Lupo

A narrator is the specific synthetic voice that reads your speaker notes aloud. You pick one each time you generate a video. The choice applies to the whole video — there's no per-slide voice override in the web UI (if you need tone variation slide-by-slide, that's what Emotions are for).

Two tiers: Standard and Premium

At the top of the voice dialog is Choose voice type:

  • Standard — high-quality neural voices, fast to render, included on all plans. Good for training, SOPs, internal enablement, product walkthroughs, and most day-to-day business use.
  • Premium — more natural prosody, wider emotional range, better at long passages. Slower to render and may cost more minutes depending on your plan. See Premium Narrators.

Switch between the two and the Choose a voice dropdown updates to show the voices in that tier. Standard voices cover every supported language; Premium voices cover fewer languages but sound more lifelike where they exist.

How to pick a voice without overthinking it

Don't A/B test voices in the abstract. Pick something plausible and generate one short test video. Three to five slides is enough to tell you if the voice fits.

A few heuristics that usually hold:

  • Training and SOPs → a voice that sounds like a helpful coworker, not a news anchor.
  • Executive summaries and stakeholder updates → steady, confident, slightly slower speed.
  • Product walkthroughs → match the audience. Internal can be casual; customer-facing needs polish.
  • Compliance or regulatory → clear and neutral. Cute is a liability when the stakes are high.
  • Marketing-adjacent enablement → Premium voices earn their keep here, because listeners notice synthetic prosody on anything that feels "produced."

Once you find a voice that works for a project, stick with it. Consistency across the course feels more professional than variety for its own sake.

Auditioning voices quickly

The fastest way to audition a voice is to pick a short, generic deck (3 slides with throwaway narration like "hello, this is slide one") and run a full video with that voice. Listening to a real clip is always more useful than reading a voice's description.

If you're building a course that will have 10 videos, spend 15 minutes on this up front. The voice you pick will be the default across everything you produce for that course.

Narration speed

Below the voice dropdown is the Narration Speed selector. Defaults to 1.0:

  • Standard voices: 0.8 to 1.3
  • Premium voices: 0.7 to 1.2

Small nudges matter. A 0.9 voice sounds more authoritative than 1.0. A 1.1 voice feels more energetic. Go beyond ±0.1 only after you've watched a full test and felt the default was off.

What to do when a voice mispronounces a word

Two options:

  1. Rewrite the word in the speaker notes so the phonetic spelling matches the pronunciation. This is ugly but effective — "Lupo" as "Loo-poh" will usually be pronounced correctly.
  2. Use an inline pronunciation hint in your notes. For advanced pronunciation control with phoneme or alias lexicons, your Lupo admin can set up a lexicon at the project level. Ask in your concierge session or email help@lupo.ai.

Where to go next

  • Emotions — vary tone within a single video using inline markers.
  • Premium Narrators — when the upgrade is worth it.
  • Translations — producing content in languages other than English.
  • Premium Tutorial — a full walkthrough of producing a video with a Premium voice.